Cybercrime detection techniques based on support vector machines

Nowadays, the users of Internet services are unthinkably increasing in the entire world. Internet services are the largest and richest source of information in the form of social networks, online applications, online videos and images, which lay among billions of web pages and blogs. The Internet users become connected to the networks for finding information, data, and friends. This situation makes suitable opportunities to hijack information and data by hackers and crooks. Cybercrimes currently is one of the most progressive offenses as problems in cyberspace, which involves any criminal act/transaction with computers and networks, such as: Spam, Drug Trafficking, Sales and Investment Fraud, Hacking, Cyber Terrorism, and Phishing [1, 2]. It is defined for the behaviours of an executable code by observing its usage dynamically. This paper presents Support Vector Machine (SVM) techniques in order to detect cybercrime in social network (Facebook) dataset.

SVM algorithm was first introduced in 1992, by Boser, Guyon, and Vapnik [3]. Support vector machines (SVMs) are a set of related supervised learning methods [4] deployed for regression and classification [5, 6]. Many software and application such as: Bioinformatics, Text Categorization, Ranking (e.g. Google search engine), Machine Vision, and Handwritten Character Recognition [7, 8], are designed using SVM algorithms in order to improve the percentage of classification accuracy. The fundamentals of SVMs have been enhanced by Vapnik [9] and became popular due to many promising features such as: better empirical performance. This method based on statistical learning theory, which is used to solve classification and regression problems [10].

The SVMs originally developed for binary classification problems, which uses the hyperplane to define separating decision boundaries among data points of different classes. “SVMs are learningalgorithms that use a hypothesis space of linear functions in a high dimensional feature space, trained with a learning algorithm from optimization theory that implements a learning bias derived from statistical learning theory” [11]. SVMs are classifying linearly/nonlinearly input space by constructing hyperplane in feature space, which separates data optimally into two categories with different classes  [10, 11].

For full text: click here

(Author: Hanif – Mohaddes Deylami, Yashwant Prasad Singh

Published by Sciedu Press)